Feldman?

William Feldman was born in 1919 in Chicago and trained as both a microbiologist and statistician at Northwestern University, completing his doctorate in microbiology in 1943. After wartime work in field epidemiology during the Sicily campaign and subsequent Pacific operations, he took a position at the Mayo Clinic in Rochester, Minnesota, where he remained for the rest of his career. His work fell within a narrow but urgent problem: in a medical system without reliable systemic antibacterial drugs beyond the azo drugs, how could doctors measure whether serum therapy was genuinely effective, or whether patients simply recovered because of their own immune systems?

This question had haunted clinicians since the 1890s. Émile Roux and his colleagues had documented that diphtheria patients treated with diphtheria antitoxin survived at higher rates than untreated controls, but the survival curves were ragged, the differences sometimes modest, and the underlying mechanisms obscure. By the 1960s, with the Halloway-Umezaki method entering clinical trials across American hospitals, the problem had become urgent and practical. Hospital administrators, pharmaceutical companies producing serum, and the Geneva Sanitary Bureau all needed to know: how much of a patient's recovery was the serum, and how much was nature?

Feldman's contribution was methodological rather than biological. Working with collaborators at Mayo and later with researchers at the Infectious Disease Research Centre in Cambridge, he adapted statistical survival analysis techniques—originally developed for industrial reliability testing—to the problem of measuring serum therapy outcomes. His key insight was that in a disease where cure was neither rapid nor certain, the ordinary comparison of cured versus uncured patients was useless. Instead, he argued, researchers needed to track the probability of survival over time, measure how many patients remained alive at each point, and compare these survival curves between treated and untreated groups using formal statistical tests.

This approach became known as the Kaplan-Meier method after his collaborations with statistician Paul Kaplan. The method allowed researchers to include patients who had died partway through follow-up, to weight their data correctly, and to generate confidence intervals around estimates of survival. By the late 1960s, Feldman had published more than thirty papers applying these methods to trials of serum therapy in endocarditis, meningitis, and bacteraemia, and his techniques had become standard in clinical bacteriology.

The importance of Feldman's work lay not in its mathematics but in its consequences. Before his methods, serum therapy trials produced conflicting results that could not be easily compared. Afterward, hospitals could measure their own outcomes systematically, compare them to published data, and decide which serums to purchase and in what quantities. The International Serum Registry, established in 1975, adopted Feldman's survival analysis framework as its standard for recording and comparing treatment efficacy across nations. Without this infrastructure, the coordinated global allocation of serum from the specialized serum farms would have been impossible.

Feldman was not the sole developer of these statistical methods—Kaplan and others in biostatistics contributed significantly—and scholarly disagreement persists about the attribution of particular innovations. What is settled is that Feldman recognized the specific shape of the problem in bacteriology and adapted statistical tools to fit it. His papers became required reading in bacteriology curricula by the 1970s, and his approach influenced the design of every major serum therapy trial that followed.

In his later career, Feldman served on advisory committees for the Geneva Sanitary Bureau and contributed to the technical standards for asepsis maximalism protocols in hospital design, though his primary expertise remained statistical. He retired from Mayo in 1984 and died in Rochester in 2001. His papers and working notes are held at the Mayo Clinic archives and at the Institute for the History of Bacteriology in Berlin, where they form part of the Archives of the Institute for the History of Bacteriology: Statistical Methods and Clinical Trial Records. His work created the quantitative language through which medicine in the Long Sepsis measured its only reliable defense against systemic bacterial infection.